No abstract

This paper presents an automatic content-based image retrieval (CBIR) system for brain tumors on T1-weighted contrast-enhanced magnetic resonance images (CE-MRI). The key challenge in CBIR systems for MR images is the semantic gap between the low-level visual information captured by the MRI machine and the high-level information perceived by the human evaluator. The traditional feature extraction methods focus only on low-level or high-level features and use some handcrafted features to reduce this gap. It is necessary to design a feature extraction framework to reduce this gap without using handcrafted features by encoding/combining low-level and high-level features. Deep learning is very powerful for feature representation that can depict low-level and high-level information completely and embed the phase of feature extraction in self-learning. Therefore, we propose a deep convolutional neural network VGG19-based novel feature extraction framework and apply closed-form metric learning to measure the similarity between the query image and database images. Furthermore, we adopt transfer learning and propose a block-wise fine-tuning strategy to enhance the retrieval performance. The extensive experiments are performed on a publicly available CE-MRI dataset that consists of three types of brain tumors (i.e., glioma, meningioma, and pituitary tumor) collected from 233 patients with a total of 3064 images across the axial, coronal, and sagittal views. Our method is more generic, as we do not use any handcrafted features; it requires minimal preprocessing, tested as robust on fivefold cross-validation, can achieve a fivefold mean average precision of 96.13%, and outperforms the state-of-the-art CBIR systems on the CE-MRI dataset. INDEX TERMS Brain tumor retrieval, block-wise fine-tuning, closed-form metric learning, convolutional neural networks, feature extraction, transfer learning.

No abstract

No abstract

No abstract

Meiotic recombination is vital for maintaining the sequence diversity in human genome. Meiosis and recombination are considered the essential phases of cell division. In meiosis, the genome is divided into equal parts for sexual reproduction whereas in recombination, the diverse genomes are combined to form new combination of genetic variations. Recombination process does not occur randomly across the genomes, it targets specific areas called recombination "hotspots" and "coldspots". Owing to huge exploration of polygenetic sequences in data banks, it is impossible to recognize the sequences through conventional methods. Looking at the significance of recombination spots, it is indispensable to develop an accurate, fast, robust, and high-throughput automated computational model. In this model, the numerical descriptors are extracted using two sequence representation schemes namely: dinucleotide composition and trinucleotide composition. The performances of seven classification algorithms were investigated. Finally, the predicted outcomes of individual classifiers are fused to form ensemble classification, which is formed through majority voting and genetic algorithm (GA). The performance of GA-based ensemble model is quite promising compared to individual classifiers and majority voting-based ensemble model. iRSpot-GAEnsC has achieved 84.46 % accuracy. The empirical results revealed that the performance of iRSpot-GAEnsC is not only higher than the examined algorithms but also better than existing methods in the literature developed so far. It is anticipated that the proposed model might be helpful for research community, academia and for drug discovery.

No abstract

No abstract